Out-of-Distribution Detection with Prototypical Outlier Proxy

TL;DR

This paper introduces Prototypical Outlier Proxy (POP), a novel framework for out-of-distribution detection that uses virtual prototypes to improve decision boundaries, achieving better accuracy and faster training and inference times.

Contribution

The paper proposes a simple framework with virtual prototypes and a hierarchical loss for OOD detection, significantly improving performance and efficiency over existing methods.

Findings

POP reduces FPR95 by up to 7.70% on CIFAR-10.

POP trains 7.2 times faster than NPOS.

POP achieves faster inference, 19.5 times quicker than NPOS.

Abstract

Out-of-distribution (OOD) detection is a crucial task for deploying deep learning models in the wild. One of the major challenges is that well-trained deep models tend to perform over-confidence on unseen test data. Recent research attempts to leverage real or synthetic outliers to mitigate the issue, which may significantly increase computational costs and be biased toward specific outlier characteristics. In this paper, we propose a simple yet effective framework, Prototypical Outlier Proxy (POP), which introduces virtual OOD prototypes to reshape the decision boundaries between ID and OOD data. Specifically, we transform the learnable classifier into a fixed one and augment it with a set of prototypical weight vectors. Then, we introduce a hierarchical similarity boundary loss to impose adaptive penalties depending on the degree of misclassification. Extensive experiments across various benchmarks demonstrate the effectiveness of POP. Notably, POP achieves average FPR95 reductions of 7.70%, 6.30%, and 5.42% over the second-best methods on CIFAR-10, CIFAR-100, and ImageNet-200, respectively. Moreover, compared to the recent method NPOS, which relies on outlier synthesis, POP trains 7.2X faster and performs inference 19.5X faster. The source code is available at: https://github.com/gmr523/pop.